Fast method for estimating the energy distribution of globular states of proteins.

By an enumeration study, we show that the energy distributions of a lattice protein sequence on all possible compact lattice configurations can be approximated by the energy distribution of shuffled sequences on a given lattice structure. We also show that the random energy model (REM) gives a good analytical approximation for the energy distribution of shuffled sequences on lattice structures. For real proteins, when a gapped threading method is used, REM calculations systematically underestimate the mean value of the energy distributions. We found that this discrepancy can be roughly compensated by a linear correction obtained from empirical fits. This result can be used to greatly reduce the computational effort in protein threading calculations.